Introduction

Despite the fact that endovascular treatment (EVT) for acute ischemic stroke is considered safe, the procedure carries a risk of intracranial hemorrhage (ICH) [1, 2]. This risk of ICH post-EVT can be attributed to mechanical lesions of the vessels that occur during EVT or reperfusion injury of brain tissue after recanalization [2, 3].

Previous studies have indicated that blood pressure (BP) may be associated with an increased risk of ICH post-EVT [2, 4]. However, these studies mainly focused on BP after, rather than during EVT. Furthermore, other studies which evaluated BP during EVT mainly assessed the relationship between decreases in BP and ICH, whereas research on the relationship between elevated BP during EVT and post-EVT ICH is sparse [5,6,7]. One study showed that BP at the time of recanalization is associated with ICH post-EVT. However, the presence of ICH was assessed on non-contrast CT between 24 and 36 h, and post-EVT elevated BP up until the non-contrast CT was not taken into account [4, 8, 9]. Moreover, assessment of ICH on non-contrast CT could lead to an overestimation of ICH due to the inability to reliably distinguish between blood and contrast [10]. An immediate post-EVT dual-energy CT (DECT) is able to overcome these limitations, as well as the ability to reliably distinguish between blood and contrast [11].

In this study, we aimed to evaluate the association between procedural BP and post-EVT ICH on DECT, in patients who underwent EVT for acute ischemic stroke.

Methods

Patient Selection

After gaining approval of the local ethics committee who waived the need for individual patient consent (METC-2020-1456), we screened all EVT records at the Maastricht University Medical Center from January 2010 up and including December 2019. Patients were considered eligible for inclusion if they had occlusion of the intracranial carotid artery (ICA), intracranial carotid artery terminus (ICA-T) or middle cerebral artery (M1/M2) and received DECT within 3 h post-EVT. Patients with pre-EVT ICH (e.g., chronic subdural hematoma or ICH due to head trauma) were excluded. Patients included in the MR CLEAN LATE trial were excluded, due to the blinded endpoint of this trial [12]. Likewise, patients receiving additional aspirin or heparin as part of MR CLEAN MED treatment allocation were excluded, since it increases the risk of ICH [13]. Patients receiving aspirin or heparin during EVT as part of regular care were not excluded.

Blood Pressure

BP measurements during EVT procedure were retrieved retrospectively from the anesthesia registration forms. As part of regular care, procedural BP generally was measured by using a noninvasive BP cuff or by using invasive BP measurements via an intra-arterial catheter and was generally recorded every three to five minutes. For our analysis, we did not apply a minimum number of measurements because this might introduce a bias by excluding patients with a very fast procedure time. In every patient, we calculated mean systolic BP (SBPmean) and mean diastolic BP (DBPmean). Mean arterial pressure (MAP) was calculated using the following formula: MAPmean = DBPmean + 1/3 (SBPmean − DBPmean) [14]. To describe BP variability, SBPmax–min was calculated subtracting the lowest SBP from the highest SBP, which has the advantage of being independent of SBPmean [15].

Anesthetic Management

During the EVT procedure, standard monitoring was applied including a 3- or 5-lead electrocardiogram, pulse oximetry, and BP monitoring. Anesthetic management consisted preferably of local anesthesia alone, but local anesthesia with procedural sedation and analgesia, or general anesthesia, could be applied up to the discretion of the responsible anesthesiologist. Generally, remifentanil and propofol were used for procedural sedation and analgesia or general anesthesia. Norepinephrine, phenylephrine, or ephedrine was used for inotropic and/or vasoactive treatment, while labetalol, clonidine, or urapidil was used for antihypertensive therapy. However, BP management during EVT was not protocolized. Therefore, hemodynamic management was also up to the discretion of the anesthesiologist. Hemodynamic interventions during EVT were retrospectively obtained from patients records, and details are presented in Supplemental Table 1.

Imaging Protocol

DECT was performed immediately, as soon as logistics allowed, post-EVT as part of regular care in our center and was performed on a second- or third-generation dual-source CT system (Somatom Definition Flash/Force, Siemens Healthcare, Forchheim, Germany). The dedicated dual-source protocol has been described elsewhere [16]. Postprocessing was performed using Syngo.via (VB40, Siemens Healthcare, Munich, Germany), which uses a three-material decomposition algorithm to differentiate between normal brain parenchyma, iodine concentration (by creating iodine overlay reconstructions), and ICH (by creating virtual post-procedural unenhanced reconstructions).

Imaging Assessment

The virtual post-procedural unenhanced reconstructions of DECT were assessed for ICH by a neuroradiologist with > 10 years of experience with DECT (A.P). ICH was scored according to the Heidelberg Criteria [17]. If a patient had both an intraparenchymal and extraparenchymal ICH, the ICH was scored as intraparenchymal, since we hypothesize that BP influences the risk of intraparenchymal ICH more than extraparenchymal ICH [18, 19]. All patients with ICH on DECT were screened for the occurrence of symptomatic ICH. Symptomatic ICH was scored according to the Heidelberg Bleeding Classification and was defined as an ICH with neurological deterioration toward admission National Institutes of Health Stroke Scale (NIHSS) of ≥ 4 points on the total NIHSS score or ≥ 2 points on any NIHSS category, or an ICH leading to surgical intervention [17]. Patients with ICH who were asymptomatic at the time of DECT, regardless of whether symptoms developed later on, were classified as asymptomatic.

Statistical Analysis

BP measurements were missing in 27% of the patients due to missing anesthesia registration forms. Data regarding clinical variables were gathered from prospective records and were overall only missing in 3%. Missing data were imputed using both single and multiple imputations by chained equations with the “mice” package (version 3.14.0) [20]. The single and multiple imputation models included the same relevant covariates. The number of imputed datasets needed to have consistent estimates of the standard error was based on the fraction of missing information (n = 23) [21].

Baseline characteristics were compared between patients without ICH and patients with intraparenchymal or extraparenchymal ICH using conventional statistics on the crude dataset.

Univariable and multivariable binary logistic regression was used to assess the association between each BP parameter and the occurrence of intraparenchymal or extraparenchymal ICH in the multiple imputed datasets. Patients with intraparenchymal ICH were excluded from the regression models analyzing BP and isolated extraparenchymal ICH, and vice versa. To avoid overfitting, the multivariable models were only adjusted for EVT duration and number of attempts during EVT, since this could be indicative for a more challenging procedure and a higher chance of complications [3, 22, 23]. However, based on univariable analysis and previous literature, four sensitivity analyses with four different pairs of adjustments were performed to investigate the robustness of the results. Additionally, an interaction effect of procedural BP with recanalization status was evaluated. Recanalization status was assessed using the expanded treatment in cerebral infarction (eTICI) score [24]. Successful recanalization was defined as eTICI ≥ 2B. All results are presented per 10 mmHg increase in BP.

In order to assess whether BP parameters differed between asymptomatic and symptomatic patients with either extraparenchymal or intraparenchymal ICH, Mann–Whitney U tests on a single imputed dataset were performed. As the time between EVT and DECT might be too short to develop a symptomatic ICH, a sensitivity analysis was performed in which patients with asymptomatic ICH at time of DECT were classified as symptomatic if neurological deterioration occurred within 24 h post-EVT. Lastly, a sensitivity analysis was performed excluding patients with missing BP measurements.

A P value < 0.05 was considered significant. Due to the exploratory nature of this study, no adjustments for multiple testing were applied. All statistical analysis were performed using R version 4.2.2.

Results

We included 478 patients in our study (Fig. 1). The median available number of procedural BP measurements per patient was 9 (IQR 6–15). Invasive BP measurements via an intra-arterial catheter only occurred in three patients. The mean of the procedural SBPmean, SBPmax, and MAPmean were 150 ± 26 mmHg, 173 ± 29 mmHg, and 101 ± 17 mmHg, respectively (Supplemental Fig. 1).

Fig. 1
figure 1

Flowchart of included patients. Abbreviations: DECT = dual-energy CT; EVT = endovascular treatment; ICA (segment) = internal carotid artery; M (segment) = middle cerebral artery

Full size image

76/478 (16%) patients showed ICH on post-EVT DECT, of which 50 (66%) were classified as extraparenchymal (of which 44 subarachnoid ICH, Supplemental Table 2) and 26 (34%) as intraparenchymal. Baseline characteristics and procedural BP of patients without ICH, patients with intraparenchymal ICH, and patients with extraparenchymal ICH are shown in Table 1. Patients with an intraparenchymal ICH more often took antiplatelet medication pre-stroke (P = 0.006) and had a higher SBPmax, SBPmean and MAPmean (P = 0.02, P = 0.002, and P = 0.006, respectively) compared to patients without ICH (Table 1). Patients with an extraparenchymal ICH had a higher number of attempts during the procedure (P < 0.001), less often achieved successful recanalization (P = 0.04), more often had a perforation as complication during EVT (P < 0.001), had longer EVT duration (P = 0.03), and less often had general anesthesia (P = 0.004) compared to patients without ICH, but no significant differences in BP parameters. Overall, there was no significant difference in frequency of IVT administration in patients with intra- or extraparenchymal hemorrhage and patients without ICH.

Table 1 Patient characteristics of patients with extra- or intraparenchymal ICH and without ICH on DECT
Full size table

Intraparenchymal and Extraparenchymal ICH

The results of multivariable regression analysis are presented in Table 2. Intraparenchymal ICH was associated with SBPmax, SBPmean, and MAPmean with adjusted odds ratios of 1.19 (95%CI 1.02–1.39), 1.22 (95%CI 1.03–1.46), and 1.40 (95%CI 1.09–1.81) per 10 mmHg increase, respectively. Extraparencymal ICH was not associated with procedural BP. Moreover, four additional models, adjusted for other significant confounders, are presented in Supplemental Tables 3–6 and did not yield different results. There was no significant interaction between BP and recanalization status on intraparenchymal or extraparenchymal ICH (Supplemental Table 7).

Table 2 Associations between blood pressure parameters and ICH
Full size table

Asymptomatic and Symptomatic ICH

Intraparenchymal ICH was symptomatic in 10 (38%) patients and extraparenchymal ICH was symptomatic in 4 (8%) patients. A complete list of asymptomatic and symptomatic ICH combinations based on the Heidelberg criteria is presented in Supplemental Table 2. BP parameters did not differ significantly between patients with an asymptomatic and symptomatic intraparenchymal or extraparenchymal ICH (Table 3). In the sensitivity analysis, three additional patients with an intraparenchymal ICH were classified as symptomatic within 24 h post-EVT, which did not yield different results (Supplemental Table 8).

Table 3 Blood pressures in patients with symptomatic and asymptomatic ICH
Full size table

Missing BP Measurements

Baseline characteristics between patients with and without missing BP measurements are presented in Supplemental Table 9. None of the baseline characteristics significantly differed between the two groups. Additionally, the sensitivity analyses excluding patients with missing BP measurements, leaving a total of 366 patients, yielded highly similar results (Supplemental Table 10).

Discussion

In this retrospective single-center study, we showed an association between procedural BP and the occurrence of immediate post-EVT intraparenchymal ICH in acute stroke patients who underwent EVT. We did not find an association with extraparenchymal ICH. While extraparenchymal ICH most commonly occurs due to (mechanical) trauma of vessels, intraparenchymal ICH occurs most commonly due to ischemic damage to the cerebral blood vessels and brain tissue and this may be influenced by BP levels.

Previous studies have demonstrated that both elevated SBP pre-EVT and elevated or variable SBP in the first 24 h post-EVT are associated with the occurrence of ICH [7, 8, 25,26,27]. However, previous literature has not demonstrated a reduction in intraparenchymal ICH rate between 24 and 36 h post-EVT by active post-procedural BP lowering [28]. Considering BP during EVT, decreases in BP have been associated with worse functional outcome and larger infarct volumes [29, 30]. The literature on elevated BP and post-EVT ICH is sparser. While we found that SBPmax, SBPmean, MAPmean were associated with a higher odds of ICH, one study found that higher procedural SBPmean and MAP were associated with a lower odds of ICH post-EVT [31]. This difference in results may be partially explained by the fact the median procedural SBPmean and MAP in their cohort (128 and 91 mmHg, respectively) were lower compared to our study cohort (150 and 101 mmHg, respectively). As their admission SBP and MAP were comparable to the characteristics of our population and previously published results from larger stroke cohorts (e.g., the MR CLEAN Registry), it is possible that more BP lowering medication was used between admission and EVT in their cohort, which may have influenced their findings [31, 32]. Still, it is possible that both elevated and decreased procedural BP have a deleterious effect on ICH development post-EVT. A potential benefit of individual BP regulation may be explored in future trials.

The frequency in which ICH was observed in our study was relatively low, namely 5.4% intraparenchymal ICH, and more specifically 2.5% parenchymal hematoma type 1 and 2 compared to 14.5–17.7% parenchymal hematoma type 1 and 2 in other procedural BP and post-EVT ICH studies [4, 31, 33]. One reason for this could be different timing of the CT scans. Even though most ICH occur within hours post-EVT it is possible that immediate post-EVT underestimates the development of ICH [34]. Another reason could be the use of non-contrast CT without dual-energy technique at 24–36 h in other studies, which can overestimate the ICH occurrence. One study evaluating patients with DECT post-EVT found 59% of ICH findings were mimicked by contrast extravasation within 18 h, and 28% of ICH findings were mimicked by contrast extravasation between 18 and 36 h, suggesting that remaining hyperdensities days after the EVT procedure could still consist of contrast extravasation rather than sole hemorrhage [35].

In our study, we did not find a significant interaction term of BP with recanalization status on ICH. Previous studies indicated that elevated procedural BP before and immediately after recanalization was associated with intraparenchymal ICH post-EVT and worse functional outcome [4, 36]. While we hypothesized that in patients who achieved successful recanalization elevated BP could cause ICH due to reperfusion injury, none of the BP parameters showed an interaction with recanalization status on the occurrence of ICH. However, due to the observational design and missing data on time point of recanalization, we could not reliably distinguish between BP before and after reperfusion. Prospective trials should evaluate this interaction further, to determine if actively lowering BP after successful recanalization could decrease occurrence of ICH and improve patient outcomes.

In our study, BP parameters did not statistically differ between asymptomatic and symptomatic ICH. Moreover, our sensitivity analysis including symptomatic ICHs at 24 h which were asymptomatic at moment of DECT did not yield different results. It is hypothesized that there are inter-individual differences in BP needed to perfuse the viable penumbra; therefore, it is possible that the same elevated procedural BP could have different outcomes (e.g., asymptomatic or symptomatic ICH) in different patients. Further randomized controlled trials could shed some light in individualized procedural BP management [37, 38].

The strengths of this study include our access to a large cohort of patients who received routinely DECT post-EVT. However, some limitations need to be considered. Since DECT post-EVT is part of standard care in our center, we retrospectively had to assess whether a patient was asymptomatic or symptomatic at time of DECT. Secondly, procedural BP were collected retrospectively, which resulted in a substantial proportion of missing BP parameters. However, our sensitivity analysis excluding patients with missing BP measurements yielded highly similar results. Third, BP is highly variable, it was only recorded every three to five minutes till the end of the procedure and only three patients had invasive intra-arterial blood pressure measurements. As a result, it is possible that valuable information in-between measurements were missed as well as BP between end of the procedure and DECT. Moreover, there was no standardized BP medication protocol during EVT. Lastly, this being a single-center study, the results might not be generalizable.

Conclusion

Procedural SBPmax, SBPmean, and MAPmean are associated with intraparenchymal ICH on immediate post-EVT DECT but not with extraparenchymal ICH. Future randomized controlled trials are needed to evaluate whether individual procedural BP management can reduce post-EVT ICH and improve clinical outcome.